The Giant Okavango and Related Ma¢C Dyke Swarms, Karoo Igneous Province, Northern Botswana

Earth and Planetary Science Letters 202 (2002) 595^606 www.elsevier.com/locate/epsl

40Ar/39Ar geochronology and structural data from the giant Okavango and related ma¢c dyke swarms, Karoo igneous province, northern Botswana

B. Le Gall a;Ã, G. Tshoso a, F. Jourdan b,G.Fe¤raud b, H. Bertrand c, J.J. Tiercelin a, A.B. Kampunzu d, M.P. Modisi d, J. Dyment a, M. Maia a

a UMR-CNRS 6538, Institut Universitaire Europe¤en de la Mer, 29280 Plouzane¤, France b UMR-CNRS 6526 Ge¤osciences Azur, Universite¤ de Nice-Sophia Antipolis, O6108 Nice, France c UMR-CNRS 5570, ENS et UCBL, 69364 Lyon, France d Department of Geology, University of Botswana, Gaborone, Botswana

Received 25 February 2002; received in revised form 11 June 2002; accepted 13 June 2002

Abstract

In NE Botswana, the Karoo dykes include a major N110‡ dyke swarm known as the Okavango giant dyke swarm (ODS/N110‡) and a second smaller set of N70‡ dykes belonging to the Sabi-Limpopo dyke swarm (SLDS/N70‡). New 40Ar/39Ar plagioclase dating of Karoo dolerites of the giant ODS/N110‡ and the SLDS/N70‡ in NE Botswana yield plateau ages between 179.6 þ 1.2 and 178.4 þ 1.1 Ma. Our data are concordant with previous 40Ar/39Ar ages for Northern Karoo dykes and lava flows exposed in western Zimbabwe. The data are tightly clustered, indicating a short-lived (179^181 Ma) flood basalt magmatism in this region. The new radiometric dates allow the definition of a diachronous Jurassic flood basalt activity in southern Africa. A significant south to north younging at the scale of the Karoo igneous province correlates with a chemical zonation from low-Ti (south) to high-Ti (north) mafic rocks. Structural measurements on the ODS/N110‡ and SLDS/N70‡ Karoo dykes of NE Botswana suggest that: (1) most of the host fractures are inherited Precambrian structures; (2) dyke emplacement occurred under unidirectional tensional stresses; (3) significant syn- and post-volcanic extensional tectonics are lacking. Combined with regional geology, these geochronological and structural data do not confirm unambiguously the triple-junction hypothesis usually put forward to support a mantle plume model for the evolution of the Karoo igneous province, prior to Gondwana breakup. ß 2002 Published by Elsevier Science B.V.

Keywords: Ar-40/Ar-39; absolute ages; dike swarms; mantle plumes; Gondwana; Karoo Basin

1. Introduction

The Karoo-Ferrar/Antarctica large igneous U 6 * Corresponding author. Tel.: +33-2-98-49-87-56; province (KFA-LIP hereafter) covers 6 10 2 Fax: +33-2-98-49-87-60. km and is one of the greatest continental £ood E-mail address: [email protected] (B. Le Gall). basalt provinces in the world (Fig. 1a). It is linked

0012-821X / 02 / $ ^ see front matter ß 2002 Published by Elsevier Science B.V. PII: S0012-821X(02)00763-X

EPSL 6302 16-9-02 596 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606

Fig. 1. Major Precambrian and Karoo structural features of West Gondwana. (a) Distribution of continental £ood basalt and related intrusive rocks of the Karoo-Ferrar/Antarctica large igneous province in a pre-drift Gondwana reconstruction. 1, exposed complexes; 2, extrapolated magmatic rocks; 3, inferred magmatic units beneath ice cover in western Antarctica. HTP and LTP, high- and low-Ti tholeiitic province, respectively. (b) Karoo tectono-magmatic framework of southern Africa. 1, magmatic complexes with 1a, £ood basalt; 1b, dykes and sills; 1c, eruptive centres; 2, sedimentary basins. C, Chilwa; L, Lesotho; LDS, Le- bombo dyke swarm; MTZ, Mozambique thinned zone; N, Nuanetsi; ODS, Okavango dyke swarm; ORDS, Olifants River dyke swarm; RRS, Rooi Rand Suite; SBDS, South Botswana dyke swarm; SLDS, Sabi-Limpopo dyke swarm; SLeDS, South Lesotho dyke swarm; SMDS, South Malawi dyke swarm; ZF, Zoetfontain fault. (c) Major basement boundaries in southern Africa. 1, palaeozoic belt; 2, proterozoic belts; 3, craton; 4, dyke swarms; 5, suture lines. EACF, East African coastal fault; GD, Great Dyke; MSZ, Magogaphate shear zone; PT, Pongola rift trend; VT, Venterdorp rift trend.

to the initial Jurassic breakup of Gondwana. The largely documented with emphasis on the petrol- African section of this LIP, the Karoo igneous ogy of igneous rocks [1], thermo-mechanical and province, consists of £ood basalts, sills and giant geotectonic evolution related to either mantle dyke swarms that together cover more than plume or subduction processes [2,3]. However, 3U106 km2 in southern Africa (Fig. 1b). The evo- limited ¢eld structural data and precise geochro- lution of the Karoo igneous province has been nological data (e.g. [4,5] have been integrated to

EPSL 6302 16-9-02 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606 597

Fig. 2. The Karoo dyke system in NE Botswana. (a) Simpli¢ed geological map showing the location of the N110‡ (Okavango) and N70‡ (Sabi-Limpopo) dyke swarms. The studied areas, including the Shashe reference section, Tsetsebjwe (Ts) and Tuli (THG), as well as isolated dated samples are shown. 1, regional envelop of the N110‡ and N70‡ dyke swarms; 2, £ood basalt; 3, Karoo basins; 4, Precambrian basement with 4a, dominantly gneissic rocks; 4b, granites; 5, major extensional faults. F, Francis- town; G, Gulubane; MF, Magogaphate fault; ODS, Okavango dyke swarm; T, Tonotha. (b) Location of dated rocks along the Shashe section. The distribution of dykes is according to published geological maps of the Geological Survey of Botswana. Loca- tion of Fig. 3 is also shown.

constrain the geotectonic models. Furthermore, interpretation: (1) the temporal relationships be- though ma¢c dyke swarms are known to be cru- tween the various dyke swarms is not yet estab- cial indicators for mantle dynamics and crustal lished, and (2) the inherited or newly formed ori- palaeostress regimes, precise geochronological gin of the involved fractures was not documented. and structural information related to the giant This paper is focused on: (1) the giant N110‡ Okavango dyke swarm are still missing. Fig. 1b Okavango dyke swarm (ODS/N110‡) which rep- indicates that the Karoo intrusive complex is resents one of the greatest ¢ssural intrusive com- dominated by four di¡erent dyke sets, namely plexes in the world [9]. It forms a 1500-km-long the Okavango (N110‡), Sabi-Limpopo (N70‡), and 100-km-wide tectono-magmatic structure ex- Lebombo (N-S) and Olifants River (N20^40‡) sys- tending from Nuanetsi in western Zimbabwe tems. These four dyke arrays converge towards through northern Botswana up to northern Na- the Nuanetsi area, in southern Zimbabwe, which mibia (Fig. 1b). (2) The N70‡ dyke swarm (SLDS/ is considered to be a rift triple-junction formed N70‡) exposed south of the ODS/N110‡ and cor- above a mantle plume [2,6^8]. However, the fol- responding to the SW tip of the Sabi-Limpopo lowing critical issues were not addressed in that dyke system [10]. Except for one 40Ar/39Ar

EPSL 6302 16-9-02 598 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606

Fig. 3. Structural features of the Karoo ODS/N110‡ swarm and its Proterozoic basement host-rocks (see location in Fig. 2b). (a) Vertical attitude of a N70‡ segmented dyke at the Shashe Dam (sample BOT/0055). (b) Small-scale segmentation of a N110‡ dyke (10 cm thick) caused by a basement-related N70‡ brittle fracture. (c) N70‡ brittle shear planes in orthogneiss from the Shashe Dam-Tonotha area. (d) N70‡ granite veins and epidote-¢lled cracks dextrally o¡set by a N110‡ (dyke-parallel) basement fault, Shashe Dam area.

whole-rock age determination on a ma¢c dyke to present ¢eld structural observations for both from the external part of the swarm [11], the dyke systems; and (3) to integrate our data in ODS/N110‡ and SLDS/N70‡ have never been the regional framework in order to generate con- dated and their structural setting never been fully strained interpretations on the rift triple-junction documented. model. The ODS/N110‡ has been variously interpreted as: (1) a failed arm of a triple rift junction above a starting mantle plume which led to the breakup 2. Geological framework and structural analysis of Gondwanaland [2,12,13]; (2) the result of a plume impinging a subducting lithosphere [3]; The Pre-Kalahari basement is well exposed in (3) a South Atlantic ocean-related magmatism NE Botswana and primarily comprises foliated emplaced along a Cretaceous continental trans- Archaean rocks of the Zimbabwe craton and verse fracture [14]. Limpopo belt and unconformably overlying Per- The objectives of this paper are: (1) to present mo-Jurassic Karoo sedimentary and volcanic se- new 40Ar/39Ar ages on plagioclases from N110‡ quences of the Central Kalahari basin and Tuli and N70‡ dyke sets in northern Botswana; (2) half-graben [15^18] (Figs. 1b and 2a).

EPSL 6302 16-9-02 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606 599

The average length of individual dykes within 3. Geochronology the ODS/N110‡ is 10 km with a mean N110‡ azi- muth parallel to the swarm trend (Fig. 2b). The 3.1. Sample description and analytical procedures highest exposed dyke density occurs in the Fran- cistown area which is drained by the Shashe Riv- Ten samples of ma¢c dykes intruding into Ar- er. The nearly unbroken exposures along the chaean gneissic granites (nine samples) and Karoo Shashe River allow a complete sampling of all sediments (one sample) in NE Botswana were se- the dykes of the ODS/N110‡ within a ca. 80- lected for 40Ar/39Ar dating. They are representa- km-long continuous section, at high angle to the tive of both the ODS/N110‡ (BOT/03-10A-17-19- swarm trend (Fig. 2b). Along the section, the 20, BOT/0028-39-55-99) and SLDS/N70‡ (BOT/ ODS/N110‡ comprises more than 170 regularly 0020) dyke sets. Note that (1) sample BOT/0055 spaced vertical dykes, 18 m thick on average is from a N70‡-trending segment of a N110‡ dyke, (140 data), with generally simple, planar, paral- and (2) the trend of the dyke BOT/0020 was not lel-sided walls (Fig. 3a). The most common devi- directly measured in the ¢eld; it was deduced ations from this simple geometry are ‘en echelon’ from a previous map of the Geological Survey apophyses indicating a sinistral shearing compo- of Botswana (Tsetsebjwe 1:50 000 sheet). The nent during dyke emplacement. sampled dykes cover a broad area and various The SLDS/N70‡ dyke set is poorly exposed to structural and geographical zones of the Karoo the SW in the Tsetsebjwe area (Fig. 2a). It de¢nes igneous complex in northern Botswana (see Fig. a ca. 50-km-wide dyke swarm which is parallel to 2 for sample location). The N110‡ dykes were the northern border fault of the Tuli half-graben sampled along the Shashe section (BOT/0099-17- (Fig. 2a). The mean thickness of the N70‡ dykes 03-10A-20-0039) and in isolated localities such as (7 data) is 35 m which is nearly twice the average the Francistown quarry (BOT/19) and the Tuli value for the N110‡ dykes measured along the basin (BOT/0028). Shashe section. All the rocks (except BOT/17) consist of ¢ne- to Since the two dyke swarms do not bene¢t from medium-grained dolerites, with intergranular/in- similar exposure conditions, the resulting discrep- tersertal texture. The dominant minerals are pla- ancy in the quality and quantity of data (140 and gioclase and augite with minor olivine and Ti- seven measured dykes for the N110‡ and N70‡ magnetite. The sample BOT/0055 contains large swarms, respectively) makes it di⁄cult to compare (up to 2 cm) phenocrysts of zoned plagioclase, the corresponding cumulative dilatations of 20% oriented parallel to the margins of the dyke. In (SLDS/N70‡) and 5% (ODS/N110‡). In both some dolerites, plagioclase and sometimes augite cases, crustal extension is exclusively accommo- form glomeroporphyritic clusters. The ground- dated by dyke injection and associated vertical mass contains fresh glass or cryptocrystalline min- jointing without concomitant extensional faulting. erals, feathery augite and skeletal plagioclase and On the other hand, a dense network of steep Ti-magnetite. The mineral phases are fresh except preexisting faults/fractures is observed within olivine which is moderately serpentinized. Some the dominantly granitoid country-rocks of both plagioclase crystals are locally cut by sericite-¢lled N110‡ and N70‡ dyke swarms. A number of these cracks. steep brittle structures are parallel to the dyke Sample BOT/17 di¡ers from other samples. It is margins. Field relationships show that ma¢c a olivine-free medium-grained gabbro made of dykes intrude along preexisting fractures (Fig. plagioclase, augite, pigeonite, Ti-magnetite, inter- 3b) that typically have various origins, i.e. stitial micropegmatite and biotite. Secondary min- strike-slip faults (Fig. 3c,d), reverse fault-related erals are more abundant than in other samples fractures, or joints. In the Tuli area, the ODS/ and consist of chlorite and sericite (interstitial N110‡ and SLDS/N70‡ dyke networks show patches or in¢lling veins in plagioclases) and ur- cross-cutting relationships that do not provide a alite after pyroxenes. de¢nite chronological template [14]. Representative major and trace element analy-

EPSL 6302 16-9-02 600 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606 ses of dated samples were performed by X-ray £uorescence spectrometry at Lyon, and are listed in Table 1 (Background Data Set1). All the dolerites, except sample BOT/17, are high-Ti tholeiites (2.1 6 TiO2 6 3.9%) containing 150^360 ppm Zr. They show chemical similarities with dykes from the southern periphery of the ODS [11] and lava £ows from the Zambezi Gorge in Zimbabwe [5]. Their chemical composition is similar to that of high-Ti Karoo igneous ma¢c rocks (Fig. 1a) which de¢ne an igneous sub-province geographi- cally centred on the study area [19,20]. The gabbro BOT/17 contains a lower (0.5%) TiO2 concentration and lower Zr (61 ppm), Nb ( 6 4 ppm) and Y(6 14 ppm) abundances, and most probably is genetically unrelated with the other ODS dykes, despite its N110‡ trend and its location within the Okavango dyke swarm (Fig. 2b). Twenty to 30 mg of fresh transparent plagioclase (grain size in the range 125^250 Wm) were separated using a Frantz magnetic separator, and then carefully selected under a binocular microscope. The samples were irradiated in the nuclear reactor at McMaster University in Hamilton, Canada, in position 5C. The total neutron £ux density during irradiation (deduced from a nom- inal value for the reactor) was 9.0U1018 ncm32. We used the Hb3gr hornblende as a £ux monitor, with an age of 1072 Ma [21]. Eleven clusters of ¢ve grains of hornblende monitor were included 40 39 in a 69-mm-long irradiation canister, and the dis- Fig. 4. Ages and ArCa/ ArK ratio spectra for plagioclase tances between them were lower than 6 mm (cor- bulk samples from the dyke swarms of NE Botswana. Ap- responding to a £ux gradient of about 0.3%). The parent ages and plateau ages are given at the 1c and 2c lev- single grains of hornblende monitor were mea- el, respectively. sured individually. The J values applied to sam- 40 39 ples were deduced from the Ar*/ ArK gradient multiplier. Argon isotopes were of the order of measured on the monitors along the canister. The 260^430, 740^4600, 800^5800 and 3^40 times the maximum error bar on the corresponding 40Ar*/ blank level for masses 40, 39, 37 and 36, respec- 39 ArK ratio is þ 0.2% in the volume where the tively. The criteria for de¢ning plateau ages were samples were included. The plagioclase bulk sam- the following: (1) at least 70% of released 39Ar; ples were step heated with a double vacuum high (2) at least three successive steps in the plateau; radiofrequency furnace, directly connected to a and (3) the integrated age (error-weighted mean) stainless steel puri¢cation line and a 120‡-12 cm of the plateau should agree with each apparent MASSE mass spectrometer working with a Bau«r- age of the plateau within a 2c con¢dence interval. Signer source and a Balzers SEV 217 electron Uncertainties on the apparent ages on each step are quoted at the 1c level (Fig. 4) and do not include the uncertainties on the age of the moni- 1 http://www.elsevier.com/locate/epsl tor. Plateau ages are given at the 2c level (Fig. 4).

EPSL 6302 16-9-02 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606 601

Fig. 5. Diagram showing age versus latitude of basalts and dolerite dykes from the Karoo igneous province. Source of 40Ar/39Ar age data: (1) Sabi-Lebombo, part of Lesotho, Transvaal and Namibia [4] (D in inset); (2) Zimbabwe [5] (J in inset); (3) Okavan- go dyke swarm (this study). U/Pb zircon ages from Lesotho are from [23] (E in inset). Only 40Ar/39Ar plagioclase ages and U/Pb zircon ages are plotted with their 2c con¢dence levels.

40 39 The uncertainties on the Ar*/ ArK ratios of the basin area (BOT/0028) (Fig. 2). The sample BOT/ monitor are included in the calculation of the 19 displays a more complicated age spectrum with plateau age uncertainty. Correction factors for concordant apparent ages at high temperature 39 37 39 interfering isotopes were ( Ar/ Ar)Ca = 7.06U (52% of Ar), corresponding to a weighted 34 36 37 34 40 39 10 ,(Ar/ Ar)Ca = 2.79U10 ,(Ar/ Ar)K = mean age of 172.6 þ 2.8 Ma. The sample BOT/ 2.97U1032. Decay constants are those of [22]. 0020 from the SLDS/N70‡ dyke swarm does not Detailed analytical data given in Table 2 can be display a plateau age but instead a £at section of obtained in the Background Data Set1 and the the age spectrum, representing 64.2% of 39Ar re- main results are compiled in Table 3 (Background leased, which indicates a weighted mean age of Data Set1). We only report age spectra data be- 178.8 þ 0.8 Ma. The existence of signi¢cantly cause the 36Ar/40Ar versus 39Ar/40Ar isochron plot higher (internally concordant) apparent ages at displays data strongly clustered near the abscissa high temperature is not clearly understood, but axis, due to atmospheric contaminations mostly could correspond to low amounts of excess argon. lower than 5%. A summary of previous and new In most cases, the low temperature fractions Karoo igneous province age data is compiled in displaying variable ages higher or lower than the Fig. 5 [4,5,23]. remaining £at part of the spectra correspond to 37 39 low and increasing ArCa/ ArK ratio, hence sug- 3.2. Results gesting that this fraction corresponds to potassic alteration phases (probably sericite ¢lling small If we exclude one high temperature apparent cracks as observed under the microscope) whereas age (representing 3% of 39Ar for the sample the higher temperature gas fractions correspond BOT/03) that is not concordant with the plateau to nearly pure plagioclase, as shown by more reg- 37 39 age, seven concordant plateau ages ranging from ular and higher ArCa/ ArK ratios. The lower 178.4 þ 1.1 to 179.3 þ 1.2 Ma (with a less precise ages obtained at low temperature clearly result age of 175.5 þ 3.6 Ma) are obtained on plagioclase from younger alteration phases. On the other from ma¢c dykes sampled along the Shashe River hand, the highest ages are more di⁄cult to ex- (BOT/03-10-20, BOT/0039-55-99) and in the Tuli plain. They could result from excess argon due

EPSL 6302 16-9-02 602 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606 to contamination of basaltic magma by the base- nous emplacement of the ODS/N110‡ and the ment host-rocks as commonly described for basal- SLDS/N70‡ Karoo dyke swarms converging at tic dykes crossing much older and K-rich host- Nuanetsi. Additional age determinations (in prog- rocks. However, in that case a clear saddle shape ress), especially on the SLDS/N70‡ dyke system, is usually observed, that is not the case here, the are required to further constrain this interpreta- high temperature apparent ages being included in tion. The lack of age di¡erences between the the plateau ages (sample BOT/0020 excepted). Al- peripheral (BOT/0099-20-0039-0028) and central ternatively, they may result from 39Ar loss from (BOT/03-10-19) dykes across the 80-km-wide tiny secondary potassium-rich mineral fractions ODS/N110‡ dyke swarm (Fig. 2b) further indi- included in the plagioclase by recoil during the cates that this latter has been emplaced either irradiation. In any case, it must be outlined that during one single short-lived igneous event or as the plagioclases from these dykes are not, or the result of successive magma injections within a slightly, a¡ected by excess argon, that is crucial very short time, i.e. within the analytical margin to obtain a precise chronology of the Karoo dyke of error of geochronological data. swarms. This could be due to the super¢cial levels The new age data presented here are similar to: where these dykes were emplaced, as attested by (1) the 40Ar/39Ar age of 178.9 þ 1.4 Ma measured the existence of Karoo lava £ows in the region on a whole-rock sample at the southern periphery (see Fig. 2), hence allowing a su⁄cient degassing of the N110‡ dyke set [11], and (2) four 40Ar/39Ar of radiogenic argon eventually incorporated into whole-rock and plagioclase ages obtained further the magma from host-rocks at deeper levels. north in Zimbabwe, on lava £ows near Victoria Sample BOT/17 shows a disturbed age spec- Falls, that range from 179.8 þ 1.2 Ma to trum characterized by a nearly £at section (but 179.2 þ 0.9 Ma (plagioclase) and from 181.3 þ 1.5 without concordant apparent ages) corresponding Ma to 179.2 þ 0.9 Ma (whole-rock and plagio- to a weighted mean age of 883 þ 4 Ma. The re- clase) [5] (Fig. 5). Thus, the 40Ar/39Ar ages for maining part of the spectrum gives higher appar- both dykes and lava £ows are tightly clustered ent ages probably caused by excess 40Ar. It is around 178^181 Ma and indicate a short time- clear that the corresponding dyke does not belong span for the bulk of the magmatic activity over to the Karoo igneous suite and is part of a much the northern Karoo igneous province of southern older Proterozoic ma¢c dyke system, though there Africa. These clustered data are from three di¡er- are no obvious ¢eld features allowing to discrim- ent laboratories using distinct monitors (FCT-3 inate the Proterozoic and Jurassic dykes in the biotite for [5], Hb3gr amphibole for this study, ODS/N110‡. and unspeci¢ed for [11]). Nevertheless, the two ages of 28.04 and 1072 Ma chosen for the two monitors FCT and Hb3gr, respectively, are in 4. Interpretation agreement with the intercalibration performed by [24]. Therefore, based on these ages and the 4.1. Timing of Karoo magmatism geographical distribution of the dated rocks, the N110‡ Okavango dyke complex, and possibly the The new 40Ar/39Ar ages presented in this paper N70‡ Sabi-Limpopo dyke swarm represent the indicate that the ODS/N110‡ and SLDS/N70‡ feeders of the high-Ti tholeiites among the North dyke sets of northern Botswana were emplaced Karoo lava pile. In the Tuli half-graben (Fig. 2a), between 179.3 þ 1.2 Ma and 178.4 þ 1.1 Ma. This N110‡ dykes intrude basaltic lavas £ows that are age range represents the best estimate for the assumed to represent the lowermost part of the ODS/N110‡, ¢rmly establishing a mid-Jurassic partly eroded Karoo £ood basalt succession. age to the giant dyke complex of NE Botswana The emplacement ages for the southern Karoo and hence excluding a Cretaceous age proposed igneous sub-province of the KFA-LIP (Fig. 5) by [14]. Although the number of dated dykes of are in the interval 185^180 Ma when only precise both directions is low, our data suggest a synchro- 40Ar=39Ar and U^Pb zircon ages are compiled

EPSL 6302 16-9-02 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606 603

[4,25,26]. Thus, taking into account the analytical the emplacement of Karoo dykes. The structural uncertainties, the Karoo igneous province was signi¢cance of this major basement fault/dyke most probably emplaced in ca. 5 Ma. At the re- zone within the Proterozoic framework of south- gional scale, the onset of Karoo magmatism ern Africa remains to be precisely determined. seems to be younger in the northern igneous The SLDS/N70‡ dyke system follows a fabric sub-province. Indeed, mineral separates (mostly which is known to represent a fundamental tec- plagioclase) 40Ar/39Ar ages from (1) Lesotho and tonic feature of the southern African sub-conti- Lebombo Karoo lava £ows; (2) Transvaal Karoo nent [29]. In NE Botswana, the northern edge of dykes and sills; and (3) central/southern Namibia the Kaapvaal craton strikes roughly N70‡ (Fig. Karoo £ood basalts display plateau ages ranging 1c), and several faults having the same trend with- from 184.7 þ 0.7 Ma to 180 þ 1.8 Ma [4], com- in this craton were rejuvenated during Karoo ex- pared to the interval 179 þ 1.2 to 178.4 þ 1.1 Ma tension, e.g. the Zoetfontain fault [30] (Fig. 1b). for Karoo ma¢c rocks in Botswana and Zim- Further north, the Magogaphate N70‡ ductile babwe (5, and this work). This N^S diachronism shear zone (Fig. 1c) also forms a major Archaean for the onset of Karoo igneous activity correlates discontinuity within the Limpopo belt [31].This with a large-scale geochemical zonation involving shear zone has been successively reactivated dur- a (younger) high-Ti basalt sub-province to the ing the Palaeoproterozoic Ubendian^Eburnean north and an (older) low-Ti basalt sub-province orogeny [31,32] and during Karoo extension to the south [19,20] (Fig. 1a). (Fig. 1b). Karoo-age basement-controlled N70‡ extensional faults also de¢ne the northern bound- 4.2. Structural and geotectonic implications ary of the Tuli half-graben [33] and extend further to the NE along the Sabi-Lebombo dyke swarm The origin of the linearity of the giant Okavan- from NE Botswana to the Mozambican coast go dyke swarm, with a constant N110‡ strike over (Fig. 1b). From the structural considerations more than 1500 km, is poorly understood, since it above, the ODS/N110‡ and SLDS/N70‡ Karoo is not concordant with any mapped basement dyke systems exposed in NE Botswana were em- structures [27]. Indeed, in NE Botswana, the placed along reactivated Precambrian fracture ODS/N110‡ aeromagnetic anomalies intersect at zones. Similar conclusions were reached by [34] high angle the NE^SW tectonic grain of the Da- for NS Karoo dykes of the Rooi Rand Suite mara/Ghanzi-Chobe Neoproterozoic belt (Fig. along the eastern boundary of the Kaapvaal cra- 1b,c) whilst across the Shashe River section, the ton (Fig. 1b,c). Thus, during Jurassic times, the N110‡ dyke array cuts sharply through com- Karoo extension rejuvenated a previously faulted plexely folded Archaean ductile fabrics (Fig. 2a). Precambrian continental crust as already pointed The 40Ar=39Ar (minimum) age of 884 þ 4 Ma out by some authors [35,36]. Once subjected to obtained on a N110‡ ma¢c dyke along the Karoo extension, concentration of extensional Shashe transect (sample BOT/17, this work) dem- strain and magmatism should have preferentially onstrates that part of the Okavango Karoo dyke occurred along preexisting discontinuities that system was emplaced along a reactivated N110‡ may fortuitously mimic a triple-junction pattern Proterozoic or older dyke/fracture zone. NW^SE- in the Nuanetsi area. The junction zones between elongated granitic bodies dated at 1022 þ 6 Ma both inherited and/or newly formed Karoo active (U^Pb SHRIMP single zircon) [28] along the faults could have yielded vertical pathways for same tectonic zone belong to this inferred Prote- Karoo magma, hence forming nodes or eruptive rozoic dislocation zone that probably extends centres within the KFA-LIP (Fig. 1b), e.g. Chil- over s 1500 km between the Kaapvaal and Zim- wa, Nuanetsi, Dufek and Beardmore Glacier [37^ babwe cratons (Fig. 1c). Therefore, the Okavango 39]. These fault-controlled feeder centres, from Karoo dyke swarm is likely to have exploited a which the dyke arrays are likely to have propa- major Precambrian crustal/lithospheric-scale dis- gated laterally, may give the appearance of plume continuity at depth that was reactivated during head impact zones. In particular, the Nuanetsi

EPSL 6302 16-9-02 604 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606 triple-junction-like fracture/dyke pattern can no vango and Sabi-Limpopo giant dyke swarms in longer be used as unequivocal evidence for recon- the Karoo igneous sub-province of NE Botswana. structing Karoo palaeostress directions and radial Our main conclusions are as follows: stress induced by a plume-related uplift (Fig. 1b). Instead, comparing the structure of the N70‡ and 1. The plateau ages for the Okavango N110‡ N110‡ dyke swarms exposed in NE Botswana dykes are clustered between 179 þ 1.2 Ma and rather suggests a unidirectional extension oriented 178.4 þ 1.1 Ma. They show that the ODS/ at N160‡ (perpendicular to the SLDS/N70‡) that N110‡ forms a major short-lived magmatic could account for (1) marked thickness variations event within the KFA-LIP, pre-dating Gond- between the N110‡ and N70‡ dyke swarms (18 m wana dispersal. The geochemical data reveal and 35 m in average, respectively) and (2) the that this dyke swarm includes high-Ti tholeiitic shearing component documented along N110‡ ma¢c magmas. dykes. 2. One (and possibly two) dyke of the SLDS/ Furthermore, an exhaustive review of Karoo N70‡ was emplaced at 178.8 þ 0.7 Ma and dyke systems in southern Africa (cf. SE Bot- thus is coeval with the ODS/N110‡ Karoo swana, S. Malawi, S. Lesotho) and Antarctica dykes. (West Dronning Maud Land) [40^42] (Fig. 1b) 3. The new ages obtained in the present work are leads to a much more complex dyke swarm dis- similar to previous ages related to dykes and tribution that does not support the plume-induced lava £ows from northern Botswana and Zim- triple-junction model. In a pre-drift reconstruc- babwe. Altogether, the geochronological data tion of Gondwanaland, the Karoo-age dykes of related to the northern Karoo igneous sub- the KFA-LIP do not converge systematically to- province indicate its emplacement between wards the inferred Nuanetsi triple-junction as- 181 and 178 Ma. sumed to occur above the KFA-LIP plume head 4. Reliable mineral ages from the southern Karoo [2,6,8]. Though a single giant plume head with igneous province in South Africa and Namibia several distinct magmatic foci structurally con- are older than the ages for the northern Karoo trolled by inherited basement features is still a igneous sub-province, suggesting that there viable model, the appropriateness of a mantle was a diachronous south to north onset of plume model requires further extensive structural Karoo magmatism in southern Africa. investigations of the Okavango, Sabi-Limpopo 5. One ma¢c dyke from the ODS/N110‡ yields a and Lebombo dyke/fracture arms, with particular minimum age of 883 þ 4 Ma; it is chemically attention paid to stress conditions that prevailed distinct (low-Ti tholeiite) from other N110‡ during dyke injection. Additionally, the fact that dykes. Thus, it shows that the Okavango giant the onset of Karoo crustal extension starts and dyke swarm includes both Proterozoic and Ju- generates rift basins more than 50 Ma before rassic dykes. the emplacement of Karoo magmatism through- 6. The role of preexisting basement brittle fabrics out southern Africa [43^44] does not support an on the emplacement of the N110‡ and N70‡ active mantle plume model for the extensional Karoo dyke swarms in NE Botswana is em- development of the Gondwana Karoo rift basins. phasized. 7. The appropriateness of previous models assign- ing the triple-junction-like pattern of Karoo 5. Conclusion dykes in Nuanetsi to a mantle plume head is questioned by our data. Integration of 40Ar=39Ar analyses of plagioclase from Karoo dolerites and geological ¢eld observations, enables us (1) to evaluate the relationship Acknowledgements between magmatism and crustal extension, and (2) to precisely address the duration of the Oka- This work is part of a partnership between the

EPSL 6302 16-9-02 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606 605

University of Botswana (Gaborone), the French [9] R.E. Ernst, K.L. Buchan, H.C. Palmer, Giant dyke Embassy in Botswana and the University of West- swarms: characteristics, distribution and geotectonic implications, in: G. Baer, A. Heimann (Eds.), Physics and ern Brittany (Brest, France). We acknowledge the Chemistry of Dykes, Balkema, Rotterdam, 1995, pp. 3^ ¢nancial support of the French Ministry of For- 21. eign A¡airs, the University of Botswana (Grant [10] K.G. Cox, R.L. Johnson, L.J. Monkman, C.J. Stillman, RPC Kaapvaal Craton Project R#442), the SU- J.R. Vail, D.M. Wood, The geology of the Nuanetsi igne- CRI 2E of the University of Western Brittany, ous province, Philos. Trans. R. Soc. London 257 (1965) 71^218. and the Universities of Nice and Lyon. P. Capiez [11] M. Elburg, A. Goldberg, Age and geochemistry of Karoo is thanked for XRF analyses. We are grateful to dykes from northeast Botswana, J. Afr. Earth Sci. 31 T. Gervais de Lafond, Head of the Cultural and (2000) 539^554. Scienti¢c Service of the French Embassy in Bot- [12] C. Reeves, A failed Gondwana spreading axis in southern swana for his support in developing this program, Africa, Nature 273 (1978) 222^223. [13] W.J. Morgan, Hotspot tracks and the opening of the At- to the Geological Survey of Botswana (Lobatse), lantic and Indian Oceans, in: C. Emeiliani (Ed.), The and to D. Ho¡man from BCL Mining, K. Callum Oceanic Lithosphere, J. Wiley, New York, 1981, pp. from Tati Mining and C. Byron from Gold Gal- 443^487. lery for their ¢eld assistance and data support. [14] C.V. Reeves, The geophysical mapping of Mesozoic dyke We acknowledge the journal’s referees Kent swarms in southern Africa and their origin in the disrup- tion of Gondwana, J. Afr. Earth Sci. 30 (2000) 499^513. Brooks and Paul Renne for their constructive re- [15] R.M. Falcon, Palynostratigraphy of the Lower Karoo Se- views. Geosciences Azur contribution No. 433. quence in the Central Sebungwe District, Mid-Zambezi [AC] basin, Rhodesia, Palaeontol. Afr. 18 (1975) 1^29. [16] R.M. Key, The geology of the area around Francistown and Phikwe, NE and Central Districts, Botswana, District References Memoir 3, Geol. Surv. Botswana, 1976. [17] D.D. Clark-Lowes, A.K. Yeats, The hydrocarbon pros- [1] A.J. Erlank, Petrogenesis of the volcanic rocks of the pects of Botswana, 1/2, Shell Coal Botswana, 1977. Karoo Province, Spec. Publ. 13, Geol. Soc. South Africa, [18] D.T. Aldiss, Pillow lavas and palynomorphs in the Karoo 1984, 395 pp. of E Botswana. Supplementary data and comment, Bot- [2] K. Burke, J.F. Dewey, Plume generated triple junctions. swana Geol. Surv. Unpubl. Rep., 10/84, 7 (1984). Key indicators in applying plate tectonics to old rocks, [19] K.G. Cox, The Karoo Province, in: J.D. MacDougall J. Geol. 81 (1972) 403^433. (Eds.), Continental Flood Basalt, Kluwer Academic Pub- [3] I.W.D. Dalziel, L.A. Lawver, J.B. Murphy, Plumes, oro- lishers, Dordrecht, 1988, pp. 239^271. genesis, and supercontinental fragmentation, Earth Plan- [20] A.J. Erlank, A.R. Duncan, J.S. Marsh, R.J. Sweeney, C.J. et. Sci. Lett. 178 (2000) 1^11. Hawkesworth, S.C. Milner, R. Miller, N.W. Rogers, A [4] R.A. Duncan, P.R. Hooper, J. Rehacek, J.S. Marsh, A.R. laterally extensive geochemical discontinuity in the sub- Duncan, The timing and duration of the Karoo igneous continental Gondwana lithosphere, in: Geochemical Evo- event, southern Gondwana, J. Geophys. Res. 102 (1997) lution of the Continental Crust, Conference abstracts, Po- 18127^18138. cos de Caldas, Brazil, 1988, pp. 1^10. [5] D.L. Jones, R.A. Duncan, J.C. Briden, D.E. Randall, C. [21] G. Turner, J.C. Huneke, F.A. Podosek, G.J. Wasserburg, MacNiocaill, Age of the Batoka basalts, northern Zim- 40Ar/39Ar ages and cosmic ray exposure ages of Appolo babwe, and the duration of Karoo Large Igneous Prov- 14 samples, Earth Planet. Sci. Lett. 12 (1971) 19^25. ince magmatism, Geochem. Geophys. Geosyst. 2 (2001) [22] R.H. Steiger, E. Jager, Subcommission on geochronology: 1^15. convention of the use of decay constants in geo- and [6] R.S. White, D.P. McKenzie, Magmatism at rift zones: the cosmochronology, Earth Planet. Sci. Lett. 36 (1977) generation of volcanic continental margins and £ood ba- 359^362. salts, J. Geophys. Res. 94 (1989) 7685^7729. [23] J. Encarnacion, T.H. Fleming, D.H. Elliot, H.V. Eales, [7] K.G. Cox, Karoo igneous activity, and the early stages of Synchronous emplacement of Ferrar and Karoo dolerites the break-up of Gondwanaland, in: B.C. Storey, T. and the early breakup of Gondwana, Geology 24 (1996) Alabaster, R.J. Pankhurst (Eds.), Magmatism and the 535^538. Causes of Continental Break-up, Geol. Soc. Spec. Publ. [24] P.R. Renne, 40Ar/39Ar age of plagioclase from Acapulco 68, Geol. Soc., London, 1992, pp. 137^148. meteorite and the problem of systematic errors in geo- [8] I.H. Campbell, R.W. Gri⁄ths, Implications of mantle chronology, Earth Planet. Sci. Lett. 175 (2000) 13^26. plume structure for the evolution of £ood basalts, Earth [25] A. Heimann, T.H. Fleming, D.H. Elliot, K.A. Foland, A Planet. Sci. Lett. 99 (1990) 79^93. short interval of Jurassic continental £ood basalt volca-

EPSL 6302 16-9-02 606 B. Le Gall et al. / Earth and Planetary Science Letters 202 (2002) 595^606

nism in Antarctica as demonstrated by 40Ar/39Ar geochro- Swarms, Geol. Assoc. Canada Spec. Pap. 34 (1987) 445^ nology, Earth Planet. Sci. Lett. 121 (1994) 19^41. 456. [26] D. Minor, S. Mukasa, Zircon U-Pb and hornblende 40Ar/ [36] R. Uken, M.K. Watkeys, An interpretation of ma¢c dyke 39Ar ages for the Dufek layered ma¢c intrusion, Antarc- swarms and their relationships with major ma¢c mag- tica: implications for the age of the Ferrar large igneous matic events on the Kaapvaal Craton and Limpopo province, Geochim. Cosmochim. Acta 61 (1997) 2497^ Belt, S. Afr. J. Geol. (1997) 341^348. 2504. [37] K.G. Cox, R.L. Jonhson, L.J. Monkman, C.J. Stillman, [27] H.V. Eales, J.S. Marsh, K.G. Cox, The Karoo igneous J.R. Vail, D.N. Wood, The geology of the Nuanetsi igne- province: an introduction, in: A.J. Erlank (Ed.), Petro- ous province, Philos. Trans. R. Soc. London A257 (1965) genesis of the Volcanic Rocks of the Karoo Province, 71^218. Geol. Soc. S. Afr. Spec. Publ. 13 (1984) 1^26. [38] A.R. Woolley, J.C. Bevan, C.J. Elliott, The Karoo doler- [28] L. VandeWel, J.M. Barton Jr., P.D. Kinny, 1.02 Ga granites of southern Malawi and their regional geochemical ite magmatism in the Tati Granite-Greenstone Terrane of implications, Miner. Mag. 43 (1979) 487^495. Botswana: implications for mineralisation and terrane [39] D.H. Elliott, Jurassic magmatism and tectonism associ- evolution, S. Afr. J. Geol. 101 (1998) 67^72. ated with Gondwanaland break-up: an Antarctic perspec- [29] M.B. Katz, East African rift and northeast lineaments: tive, in: B.C. Storey, T. Alabaster, R.J. Pankhurst (Eds.), continental spreading-transform system?, J. Afr. Earth Magmatism and the Causes of Continental Break-up, Sci. 8 (1987) 103^107. Geol. Soc. Spec. Publ. 68, Geol. Soc. London, 1992, pp. [30] D.G. Hutchins, C.V. Reeves, Regional geophysical explo- 165^184. ration of the Kalahari in Botswana, Tectonophysics 69 [40] F. Walker, A. Poldeervaart, Karroo dolerites of the (1980) 201^220. Union of South Africa, Bull. Geol. Soc. Am. 60 (1949) [31] S. McCourt, R.A. Armstrong, SHRIMP U-Pb zircon geo- 591^706. chronology of granites from the Central Zone, Limpopo [41] R. MacDonald, R. Crossley, K.S. Waterhouse, Karoo belt, southern Africa: implications for the age of the Lim- basalts of southern Malawi and their regional petrogene- popo Orogeny, S. Afr. Geol. Soc. 101 (1998) 329^338. sis signi¢cance, Miner. Mag. 47 (1983) 281^289. [32] L. Holzer, J.M. Barton, B.K. Paya, J.D. Kramers, Tecto- [42] G. Spaeth, P. Schull, A survey of Mesozoic dolerite dikes nothermal history of the western part of the Limpopo from Western Neuschwabenland, Antarctica, and their belt: tectonic models and new perspectives, J. Afr. Earth geotectonic signi¢cance, Polarforschung 57 (1987) 93^113. Sci. 28 (1999) 383^402. [43] O. Catuneanu, P.J. Hancox, B.S. Rubidge, Reciprocal [33] J.D. Bennett, The tectonometamorphic complex of East- £exural behaviour and contrasting stratigraphies: a new ern Botswana (1:500 000 scale map), Govt. Printer, Ga- basin development model for the Karoo retroarc foreland borone, 1971. system, South Africa, Basin Res. 10 (1998) 417^439. [34] J.W. Bristow, The geochronology and geochemistry of [44] J.L. Orpen, C.J. Swain, C. Nugent, P.P. Zhou, Wrench- Karoo volcanics in the Lebombo and adjacent areas, fault and half-graben tectonics in the development of the PhD thesis, University of Cape Town, 1980. Palaeozoic Zambezi Karoo basins in Zimbabwe and re- [35] D.R. Hunter, D.L. Reid, Ma¢c dyke swarms in southern gional implications, J. Afr. Earth Sci. 8 (2/3) (1989) 215^ Africa, in: H.C. Halls, W.F. Fahrig (Eds.), Ma¢c Dyke 229.

EPSL 6302 16-9-02